Académique Documents
Professionnel Documents
Culture Documents
=
2 1
2 1
2
1
N N
S
X X
t
s
s
s
The degrees of freedom v associated with this variance estimate is also
approximated using the Welch-Satterthwaite equation:
4
See Appendix A
38
2
2
2
4
2
1
2
1
4
1
2
2
2
2
1
. .
2 1
v N
s
v N
s
N
s
N
s
v
+
+
=
The statistics are then used with the t-distribution to test the null
hypothesis that the two means are equal.
39
9 INFORMATION CONTENT OF CREDIT WATCH
PLACEMENT AND BOND RATING CHANGES
9.1 Stock Related CAR for (-1, +1) for Entire Sample
In order to determine whether credit watch emplacements as well as long
term rating changes are informative events, we examine stock related CAR for a
3 day (-1, +1) window around the events using a standard event study
methodology.
Culmulative Abnormal Returns (CARs) are calculated as the cumulative
difference between the daily raw stock return and the concurrent local market
index for each of the 42 countries in our sample, as defined in DataStream. As a
robustness check, we also repeat all analyses using an alternative window period
of (-3, +3). The choice of event window does not appear to alter the significance
of our results.
40
Event Type
Avg Stock Related
CAR in (-1, +1)
t-statistic
# of
observations
Negative Credit Watch -1.37% -7.15 2,423
Positive Credit Watch 0.13% 0.90 864
Long Term Rating Downgrade -1.33% -7.11 2,435
Long Term Rating Upgrade 0.07% 0.75 1,604
Total: 7,326
Table 4
Average Stock Related Culmulative Abnormal Return (CAR) for 3 day period in window (-1, +1) centered around the
credit event at day 0 is calculated as the difference between the daily raw stock return and the concurrent local market
index (as defined in DataStream).
Average Stock Related CAR for (-1, +1) for Entire Sample for Positive / Negative Credit
Watch and Long Term Rating Upgrade / Downgrades
Fig 2: Stock Related CAR for Entire Sample
Stock Related CAR for Entire Sample
-1.6
-1.4
-1.2
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
Positive Credit Watch Negative Credit Watch Long Term Rating Upgrade Long Term Rating Downgrade
S
t
o
c
k
R
e
l
a
t
e
d
C
A
R
i
n
%
In Table 4 (also show in Fig 2), we consider four subsets of our sample.
These are positive / negative credit watch placements, and long term bond rating
upgrades / downgrades. If bond rating agencies are able to provide new
information to investors through credit watch placements or long term bond rating
41
events, then we should observe a significant stock-related CAR corresponding to
the credit watches / rating events. We find that stock-related CAR to negative
credit watch and bond rating downgrades are statistically significant. CAR
associated with negative credit watch is statistically significant at 1.37% (t-
statistic of 7.15). CAR associated with bond rating downgrades are also
statistically significant at 1.33% (t-statistic of 7.11).
Conversely CAR associated with upgrades are generally not significant.
There is only a 0.13% CAR associated with positive credit watch, and a 0.07 %
CAR associated with bond rating upgrades (both with t-statistics that are not
significant at the 2.5% level of significance). Our findings are consistent with the
bulk of academic literature on bond rating changes, e.g. Hand, Holthausen &
Leftwich (1992), Goh & Ederington (1993 and 1998) and Hite & Warga (1997),
which find that on the whole, bond rating downgrades are significant, while bond
rating upgrades do not result in a significant price reaction, and therefore by
extension, are not informative. One possible explanation for this is that firms
tend to disseminate good news aggressively, while withholding bad news; hence,
a bond rating downgrade provides more new information to investors (see Goh &
Ederington, 1993).
42
9.2 Stock Related CAR for (-1, +1) by Economy Classification
Avg Stock Related
CAR in (-1, +1)
t-statistic
# of
observations
Negative Credit Watch
(Developed Markets)
-1.44% -6.79 2,105
Negative Credit Watch
(Emerging Markets)
-0.88% -2.27 318
Welch t-test for difference
in sample means
t-statistic = 1.2521
df = 524
Positive Credit Watch
(Developed Markets)
0.10% 0.70 770
Positive Credit Watch
(Emerging Markets)
0.30% 0.76 94
Welch t-test for difference
in sample means
t-statistic = 0.4592
df = 121
Total: 3,287
Table 5
Average Stock Related CAR for (-1, +1) for Entire Sample of Positive / Negative Credit
Watches partitioned by Classification of Country of Domicile
Average Stock Related Culmulative Abnormal Return (CAR) for 3 day period in window (-1, +1) centered around the
credit event at day 0 is calculated as the difference between the daily raw stock return and the concurrent local market
index (as defined in DataStream).
43
Fig 3: Stock Related CAR for Credit Watch by Classification of Economy
Stock Related CAR for Credit Watch
-1.60%
-1.40%
-1.20%
-1.00%
-0.80%
-0.60%
-0.40%
-0.20%
0.00%
0.20%
0.40%
Positive Credit Watch
(Developed Markets)
Positive Credit Watch
(Emerging Markets)
Negative Credit Watch
(Developed Markets)
Negative Credit Watch
(Emerging Markets)
%
C
A
R
Comparing market reactions to negative credit watch emplacement and
bond rating downgrades between developed markets and emerging markets in
Table 5, we find that developed markets generally exhibit a larger reaction to
negative credit watch and bond rating downgrades compared to emerging
markets. Developed markets have an average stock-related CAR of -1.44% (t-
statistic of 6.79) for negative credit watch emplacements, compared to only -
0.88% (t-statistic of 2.27) for emerging markets. Additionally, developed
44
markets have an average stock related CAR of -1.47% (t-statistic of 6.84) for
bond rating downgrades, compared to only -0.76% (t-statistic of 2.08) for
emerging markets (Table 6). Results for positive credit watch / upgrades when
segregated by classification of economy are not significant. These results
indicate that rating agencies announcements for companies domiciled in
developed markets carry greater new information content compared to
announcements related to companies domiciled in emerging markets. This could
be due to greater information leakage in emerging markets due to fewer
restrictions against insider trading, etc. We investigate preannouncement trading
effects as a proxy of insider trading in subsequent sections.
45
Avg Stock Related
CAR in (-1, +1)
t-statistic
# of
observations
Long Term Rating Downgrade
(Developed Markets)
-1.47% -6.836782681 1,954
Long Term Rating Downgrade
(Emerging Markets)
-0.76% -2.083303349 481
Welch t-test for difference
in sample means
t-statistic = 1.6853
df = 848
Long Term Rating Upgrade
(Developed Markets)
0.08% 0.774759269 1,205
Long Term Rating Upgrade
(Emerging Markets)
0.03% 0.147906626 399
Welch t-test for difference
in sample means
t-statistic = 0.2644
df = 690
Total: 4,039
Table 6
Average Stock Related CAR for (-1, +1) for Entire Sample of Long Term Rating Upgrades /
Downgrades partitioned by Classification of Country of Domicile
Average Stock Related Culmulative Abnormal Return (CAR) for 3 day period in window (-1, +1) centered around the
credit event at day 0 is calculated as the difference between the daily raw stock return and the concurrent local market
index (as defined in DataStream).
46
Fig 4: Stock Related CAR for Rating Change by Classification of Economy
Stock Related CAR for Rating Change
-1.60%
-1.40%
-1.20%
-1.00%
-0.80%
-0.60%
-0.40%
-0.20%
0.00%
0.20%
Long Term Rating
Upgrade (Developed
Markets)
Long Term Rating
Upgrade (Emerging
Markets)
Long Term Rating
Downgrade (Developed
Markets)
Long Term Rating
Downgrade (Emerging
Markets)
%
C
A
R
47
9.3 Stock Related CAR for (-1, +1) by Linkage between CW and RC
Table 7 and 8 reports results that are additionally segregated by
informative / uninformative credit watch, and expected / unexpected bond rating
changes. Our results indicate that being put on credit watch is an effective tool to
reduce stock price volatility around actual bond rating changes. In developed
markets, the stock related CAR surrounding a surprise long term rating
downgrade is -2.09%, compared to -0.85% in the case of an expected long term
rating downgrade. Similarly, for emerging markets, the stock related CAR
surrounding a surprise long term rating downgrade is -0.85%, compared to -
0.49% for expected long term rating downgrades (t-statistics for the emerging
market average reactions are not significant at the 2.5% level of significance).
The maller difference for emerging markets could be that information leakage
has reduced the informational advantage that unexpected downgrades have over
expected downgrades.
48
Event Type
Average Stock Related CAR
(t-statistics in parantheses)
# of observations
Surprise Long Term Downgrades
-1.76%
(-5.97)
1,347
Expected Long Term Downgrades
-0.81%
(-3.89)
1,088
Welch t-test for difference in
sample means
t-statistic = 2.6304
df = 2312
Surprise Long Term Upgrades
0.02%
(0.18)
1,155
Expected Long Term Upgrades
0.19%
(1.28)
449
Welch t-test for difference in
sample means
t-statistic = 0.9277
df = 974
Surprise Long Term Downgrades
(Developed Markets)
-2.09%
(-5.74)
983
Expected Long Term Downgrades
(Developed Markets)
-0.85%
(-3.75)
971
Welch t-test for difference in
sample means
t-statistic = 2.9038
df = 1636
Surprise Long Term Downgrades
(Emerging Markets)
-0.85%
(-1.85)
364
Expected Long Term Downgrades
(Emerging Markets)
-0.49%
(-1.06)
117
Welch t-test for difference in
sample means
t-statistic = 0.5355
df = 343
Surprise Long Term Upgrades
(Developed Markets)
0.12%
(0.92)
824
Expected Long Term Upgrades
(Developed Markets)
-0.01%
(-0.08)
381
Welch t-test for difference in
sample means
t-statistic = 0.6693
df = 950
Surprise Long Term Upgrades
(Emerging Markets)
-0.24%
(-1.26)
331
Expected Long Term Upgrades
(Emerging Markets)
1.33%
(2.86)
68
Welch t-test for difference in
sample means
t-statistic = 3.1209
df = 91
Total: 4,039
Average Stock Related CAR for 3 day period in window (-1, +1) centered around long term rating event on
day 0 partitioned by whether the rating event was preceded by a credit watch or not in the correct direction
within a year.
Average Stock Related CAR for Entire Sample, Partitioned by whether Rating
Change is Surprised or Expected (i.e. preceded by corresponding credit watch)
Table 7
Panel C: Upgrades by Market Type
Panel A: Overall Sample
Panel B: Downgrades by Market Type
49
Event Type
Average Stock Related CAR
(t-statistics in parantheses)
# of observations
Informative Negative Credit Watch
-1.81%
(-6.61)
1,088
Uninformative Negative Credit Watch
-1.00%
(-3.79)
1,335
Welch t-test for difference in
sample means
t-statistic = 2.1183
df = 2377
Informative Positive Credit Watch
0.14%
(0.75)
449
Uninformative Positive Credit Watch
0.11%
(0.52)
415
Welch t-test for difference in
sample means
t-statistic = 0.1079
df = 840
Informative Negative Credit Watch
(Developed Markets)
-1.82%
(-6.15)
972
Uninformative Negative Credit Watch
(Developed Markets)
-1.11%
(-3.71)
1,133
Welch t-test for difference in
sample means
t-statistic = 1.6661
df = 2096
Informative Negative Credit Watch
(Emerging Markets)
-1.76%
(-2.55)
116
Uninformative Negative Credit Watch
(Emerging Markets)
-0.38%
(-0.81)
202
Welch t-test for difference in
sample means
t-statistic = 1.6524
df = 217
Informative Positive Credit Watch
(Developed Markets)
0.08%
(0.38)
381
Uninformative Positive Credit Watch
(Developed Markets)
0.13%
(0.60)
389
Welch t-test for difference in
sample means
t-statistic = 0.1806
df = 763
Informative Positive Credit Watch
(Emerging Markets)
0.49%
(1.02)
68
Uninformative Positive Credit Watch
(Emerging Markets)
-0.21%
(-0.32)
26
Welch t-test for difference in
sample means
t-statistic = 0.8571
df = 53
Total: 3,287
Panel C: Positive Watch
Panel B: Negative Watch
Table 8
Average Stock Related CAR for Entire Sample, Partitioned by whether Credit
Watch was Informative or Uninformative
Average Stock Related CAR for 3 day period in window (-1, +1) centered around credit watch on day 0
partitioned by whether the credit watch was resolved by a rating event or not in the correct direction within
a year.
Panel A: Overall Sample
50
Fig 5: Stock Related CAR for Credit Watch by Informativeness
Stock Related CAR for Credit Watch
-2
-1.5
-1
-0.5
0
0.5
1
I
n
f
o
r
m
a
t
i
v
e
P
o
s
i
t
i
v
e
(
D
e
v
e
l
o
p
e
d
)
I
n
f
o
r
m
a
t
i
v
e
P
o
s
i
t
i
v
e
(
E
m
e
r
g
i
n
g
)
U
n
i
n
f
o
r
m
a
t
i
v
e
P
o
s
i
t
i
v
e
(
D
e
v
e
l
o
p
e
d
)
U
n
i
n
f
o
r
m
a
t
i
v
e
P
o
s
i
t
i
v
e
(
E
m
e
r
g
i
n
g
)
I
n
f
o
r
m
a
t
i
v
e
N
e
g
a
t
i
v
e
(
D
e
v
e
l
o
p
e
d
)
I
n
f
o
r
m
a
t
i
v
e
N
e
g
a
t
i
v
e
(
E
m
e
r
g
i
n
g
)
U
n
i
n
f
o
r
m
a
t
i
v
e
N
e
g
a
t
i
v
e
(
D
e
v
e
l
o
p
e
d
)
U
n
i
n
f
o
r
m
a
t
i
v
e
N
e
g
a
t
i
v
e
(
E
m
e
r
g
i
n
g
)
%
C
A
R
51
Fig 6: Stock Related CAR for Rating Change by Linkage with CW
Stock Related CAR for Rating Changes
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
1.5
S
u
r
p
r
i
s
e
U
p
g
r
a
d
e
s
(
D
e
v
e
l
o
p
e
d
)
S
u
r
p
r
i
s
e
U
p
g
r
a
d
e
s
(
E
m
e
r
g
i
n
g
)
E
x
p
e
c
t
e
d
U
p
g
r
a
d
e
s
(
D
e
v
e
l
o
p
e
d
)
E
x
p
e
c
t
e
d
U
p
g
r
a
d
e
s
(
E
m
e
r
g
i
n
g
)
S
u
r
p
r
i
s
e
D
o
w
n
g
r
a
d
e
s
(
D
e
v
e
l
o
p
e
d
)
S
u
r
p
r
i
s
e
D
o
w
n
g
r
a
d
e
s
(
E
m
e
r
g
i
n
g
)
E
x
p
e
c
t
e
d
D
o
w
n
g
r
a
d
e
s
(
D
e
v
e
l
o
p
e
d
)
E
x
p
e
c
t
e
d
D
o
w
n
g
r
a
d
e
s
(
E
m
e
r
g
i
n
g
)
%
C
A
R
We note that the results for upgrades, segregated by expected or
unexpected are also not significant, except for the case of expected upgrades in
emerging markets. In developed markets, the stock related CAR surrounding an
expected long term rating upgrade is -0.01%, compared to 0.12% in the case of
an unexpected long term rating upgrade (t-statistics not significant at 2.5% level).
Similarly, for emerging markets, the stock related CAR surrounding an expected
long term rating upgrade is 1.33% (t-statistic of 2.86 is significant), compared to -
0.24% for unexpected long term rating upgrades. We note that the significant
52
results for expected upgrades in emerging markets may be due to the fact that a
majority of events in this category occur when the country MSCI indices exhibit
strongly positive returns. The average MSCI country index returns during the
month of expected upgrade events in emerging markets is 2.23%, compared to
1.13% on average for upgrades in general. Cross sectional regression analysis
(Table 13) shows that the state of the MSCI country index (as a proxy for the
macroeconomic environment in the corporations country of domicile) is
significant in explaining variation in reactions to upgrades in general. However,
the same variable is not significant in explaining variation in reactions to
downgrades. This could be because investors weigh the strength of a
corporations macroeconomic environment in tempering expectations of future
positive prospects by rating agencies.
53
9.4 Stock Related CAR for (-1, +1) by Rating Band Transitions
Event Type
Average Stock Related
CAR
(t-statistics in
parantheses)
# of observations
A1: Downgrades within Investment Grade
-0.30%
(-2.22)
1,297
A2: Downgrades to Speculative Grade
-2.01%
(-2.81)
271
A3: Downgrades within Speculative Grade
-2.67%
(-6.23)
867
Welch t-test for difference in sample means
between sample A1 and sample A2
t-statistic = 2.3543
df = 289
Welch t-test for difference in sample means
between sample A2 and sample A3
t-statistic = 0.7872
df = 478
A4: Upgrades within Investment Grade
-0.11%
(-1.30)
843
A5:Upgrades to Investment Grade
0.23%
(1.32)
222
A6: Upgrades within Speculative Grade
0.28%
(1.25)
539
Welch t-test for difference in sample means
between sample A4 and sample A5
t-statistic = 1.7542
df = 333
Welch t-test for difference in sample means
between sample A5 and sample A6
t-statistic = 0.1760
df = 732
Total: 4,039
B1: Downgrades within Investment Grade
-0.36%
(-2.66)
1,213
B2: Downgrades to Speculative Grade
-2.38%
(-2.63)
207
B3: Downgrades within Speculative Grade
-3.65%
(-5.84)
534
Welch t-test for difference in sample means
between sample B1 and sample B2
t-statistic = 2.2084
df = 215
Welch t-test for difference in sample means
between sample B2 and sample B3
t-statistic = 1.1577
df = 413
B4: Upgrades within Investment Grade
-0.07%
(-0.85)
747
B5: Upgrades to Investment Grade
0.02%
(0.08)
153
B6: Upgrades within Speculative Grade
0.50%
(1.48)
305
Welch t-test for difference in sample means
between sample B4 and sample B5
t-statistic = 0.4163
df = 216
Welch t-test for difference in sample means
between sample B5 and sample B6
t-statistic = 1.2333
df = 445
Total: 3,159
C1: Downgrades within Investment Grade
0.57%
(0.80)
84
C2: Downgrades to Speculative Grade
-0.82%
(-1.07)
64
C3: Downgrades within Speculative Grade
-1.08%
(-2.29)
333
Welch t-test for difference in sample means
between sample C1 and sample C2
t-statistic = 1.3336
df = 139
Welch t-test for difference in sample means
between sample C2 and sample C3
t-statistic = 0.2973
df = 117
C4: Upgrades within Investment Grade
-0.38%
(-1.42)
96
C5: Upgrades to Investment Grade
0.70%
(2.11)
69
C6: Upgrades within Speculative Grade
-0.01%
(-0.03)
234
Welch t-test for difference in sample means
between sample C4 and sample C5
t-statistic = 2.5315
df = 141
Welch t-test for difference in sample means
between sample C5 and sample C6
t-statistic = 1.6614
df = 163
Total: 880
Table 9
Panel B: Developed Markets
Panel C: Emerging Markets
Panel A: Overall Sample
Average Stock Related CAR for Long Term Rating Events for 3 day period in window (-1, +1) centered around
day 0 partitioned by whether the rating event results in a transition from investment grade to non-investment
grade (for downgrades), and vice versa for upgrades
Average Stock Related CAR for Long Term Rating Events, Partitioned by whether
Event results in an Investment Grade Transition
54
Fig 7: Stock Related CAR for Rating Change by Presence of Investment Grade
Transitions
Investment Grade Transitions vs Non Investment Grade Transitions
-3
-2
-2
-1
-1
0
1
Non Investment Grade Transition
Upgrades
Investment Grade Transition
Upgrades
Non Investment Grade Transition
Downgrades
Investment Grade Transition
Downgrades
%
C
A
R
Table 9 shows that rating downgrades which result in a change in rating
bands (i.e. from investment grade to non-investment grade) result in a stronger
stock-related CAR compared to rating changes. In developed markets, rating
downgrades that result in a change in rating bands (from investment grade to
non-investment grade) have an average stock-related CAR of -2.38%, compared
to -0.36% for rating downgrades that stay within the investment grade band (t-
statistics are 2.63 and 2.66 respectively). In emerging markets, rating
downgrades that result in a change to non-investment grade have an average
stock-related CAR of -0.82%, compared to 0.57% for rating downgrades that stay
55
within the investment grade band we note that for emerging markets, neither t-
statistic is significant at the 2.5% level of significance.
We note that these results are consistent with findings from US markets,
which show that downgrades within speculative grade result in the largest stock
related CAR compared to downgrades within investment grade, and from
investment grade to speculative grade. This is because the largest increases in
required yield to maturity of the bonds result from downgrades within the
speculative grade.
In developed markets, rating upgrades that result in a change in rating
bands have an average stock-related CAR of 0.02%, compared to -0.07% for
rating upgrades that stay within the investment grade band (both t-statistics are
not significant). In emerging markets, rating upgrades that result in a change in
rating bands have an average stock-related CAR of 0.70%, compared to -0.38%
for rating upgrades that stay within the investment grade band (the t-statistic for
upgrades that cross the investment grade band in emerging markets is
significant).
56
9.5 Stock Related CAR for (-1, +1) by State of Local Market
Event Type
Average Stock Related
CAR
(t-statistics in
parantheses)
# of observations
Downgrades in Down Local Market
-1.54%
(-4.92)
1,120
Downgrades in Up Local Market
-1.16%
(-5.20)
1,315
Welch t-test for difference in sample means
t-statistic = 0.9855
df = 2085
Upgrades in Down Local Market
0.01%
(-0.09)
619
Upgrades in Up Local Market
0.12%
(1.17)
985
Welch t-test for difference in sample means
t-statistic = 0.6816
df = 1067
Total: 4,039
Downgrades in Down Local Market -2.00%
(-5.26)
841
Downgrades in Up Local Market -1.08%
(-4.38)
1,113
Welch t-test for difference in sample means
t-statistic = 2.0284
df = 1496
Upgrades in Down Local Market 0.04%
(0.19)
464
Upgrades in Up Local Market 0.11%
(0.99)
741
Welch t-test for difference in sample means
t-statistic = 0.2865
df = 712
Total: 3,159
Downgrades in Down Local Market -0.15%
(-0.30)
279
Downgrades in Up Local Market -1.60%
(-3.14)
202
Welch t-test for difference in sample means
t-statistic = 2.0115
df = 466
Upgrades in Down Local Market -0.18%
(-0.73)
155
Upgrades in Up Local Market 0.16%
(0.63)
244
Welch t-test for difference in sample means
t-statistic = 0.9613
df = 381
Total: 880
Table 10
Average Stock Related CAR for Long Term Rating Events, Partitioned by whether
Event occurs in an "Up' local market, or "Down" local market
Average Stock Related CAR for long term rating events for 3 day period in window (-1, +1) centered around day 0
partitioned by whether the event occurs during an "up" local market or "down" local market
Panel A: Overall Sample
Panel B: Developed Markets
Panel C: Emerging Markets
57
Event Type
Average Stock Related
CAR
(t-statistics in
parantheses)
# of observations
Negative Credit Watch in Down Local Market
-1.95%
(-5.90)
1,144
Negative Credit Watch in Up Local Market
-0.84%
(-4.06)
1,279
Welch t-test for difference in sample means
t-statistic = 2.8456
df = 1948
Positive Credit Watch in Down Local Market
-0.16%
(0.67)
331
Postive Credit Watch in Up Local Market
0.10%
(0.61)
533
Welch t-test for difference in sample means
t-statistic = 0.1939
df = 640
Total: 3,287
Negative Credit Watch in Down Local Market
(Developed Markets)
-2.19%
(-5.75) 959
Negative Credit Watch in Up Local Market
(Developed Markets)
-0.81%
(-3.66)
1146
Welch t-test for difference in sample means
t-statistic = 3.1168
df = 1572
Positive Credit Watch in Down Local Market
(Developed Markets)
0.17%
(0.66)
289
Positive Credit Watch in Up Local Market
(Developed Markets)
0.06%
(0.36) 481
Welch t-test for difference in sample means
t-statistic = 0.3412
df = 560
Total: 2,875
Negative Credit Watch in Down Local Market
(Emerging Markets)
-0.74%
(-1.37) 185
Negative Credit Watch in Up Local Market
(Emerging Markets)
-1.08%
(-1.97)
133
Welch t-test for difference in sample means
t-statistic = 0.4416
df = 306
Positive Credit Watch in Down Local Market
(Emerging Markets)
0.09%
(0.13)
42
Positive Credit Watch in Up Local Market
(Emerging Markets)
0.47%
(1.02) 52
Welch t-test for difference in sample means
t-statistic = 0.4709
df = 75
Total: 412
Table 11
Average Stock Related CAR for Credit Watch Events, Partitioned by whether Event
occurs in an "Up' local market, or "Down" local market
Average Stock Related CAR for credit watch events for 3 day period in window (-1, +1) centered around day 0
partitioned by whether the event occurs during an "up" local market or "down" local market
Panel A: Overall Sample
Panel B: Developed Markets
Panel C: Emerging Markets
58
Fig 8: Stock Related CAR by State of the Local Market Index
Stock Related CAR by State of Market
-2.5
-2
-1.5
-1
-0.5
0
0.5
P
o
s
i
t
i
v
e
C
r
e
d
i
t
W
a
t
c
h
i
n
D
o
w
n
L
o
c
a
l
M
a
r
k
e
t
P
o
s
t
i
v
e
C
r
e
d
i
t
W
a
t
c
h
i
n
U
p
L
o
c
a
l
M
a
r
k
e
t
N
e
g
a
t
i
v
e
C
r
e
d
i
t
W
a
t
c
h
i
n
D
o
w
n
L
o
c
a
l
M
a
r
k
e
t
N
e
g
a
t
i
v
e
C
r
e
d
i
t
W
a
t
c
h
i
n
U
p
L
o
c
a
l
M
a
r
k
e
t
U
p
g
r
a
d
e
s
i
n
D
o
w
n
L
o
c
a
l
M
a
r
k
e
t
U
p
g
r
a
d
e
s
i
n
U
p
L
o
c
a
l
M
a
r
k
e
t
D
o
w
n
g
r
a
d
e
s
i
n
D
o
w
n
L
o
c
a
l
M
a
r
k
e
t
D
o
w
n
g
r
a
d
e
s
i
n
U
p
L
o
c
a
l
M
a
r
k
e
t
%
C
A
R
Additionally, Table 10 & 11 shows that rating downgrades in developed
markets that occur during periods where the local MSCI country index is up
exhibit less negative returns. Rating downgrades in developed markets that
occur when the local MSCI country index is up have an average stock-related
CAR of -1.08%, compared to -2.00% when the index is down. The picture is not
as clear in emerging markets. Rating downgrades in emerging markets that
occur when the local MSCI country index is up have an average stock-related
CAR of -1.60%. As the average stock-related CAR of -0.15% when the index is
down is not significant (t-statistic of -0.30), there is no basis for making the same
59
comparison as with developed markets. As seen in table 10, none of the t-
statistics for the average stock-related CAR with upgrades are significant.
Negative Credit Watch in developed markets also exhibit a less negative stock
related CAR when the local MSCI index is up. Stock related CAR is 1.95% in a
down market, compared to 0.84% in an up market. All t-statistics are significant
at the 2.5% level of significance. For emerging markets, the stock related CAR
to negative credit watch in a down local market is not significant at the 2.5% level
of significance, so there is no basis for comparison.
60
10. PREANNOUNCEMENT TRADING EFFECTS
We note that there is some evidence of insider trading in both developed
and emerging markets for long term downgrades, with a greater
preannouncement effect in the form of negative abnormal returns in emerging
markets. Table 12 shows that in the -50 to -26 day window before a long term
ratings downgrade, developed (emerging) markets exhibit an average stock
related CAR of -2.35% (-2.79%). Both sets of averages have significant t-
statistics of -9.04 (-3.35). However, in the -25 to -1 day window before the
downgrade, emerging markets exhibit a much larger -3.42% abnormal CAR,
compared to only -1.84% for developed markets. Once again, both sets of t-
statistics are significant at -7.08 (-3.69).
The larger preannouncement reaction for emerging markets is also
verified by the Welch t-test, which has a t-statistic of 1.64 for the difference
between preannouncement reactions in the 2 class of markets in the -25 to -1
window. We also note that all the t-statistics for the preannouncement
announcements in (-50, -26) and (-25, -1) for both markets are significant. For
negative credit watch, we note that there is no basis for comparison of magnitude
61
of insider trading because the t-statistics for preannouncement effects in
emerging markets are all not significant.
62
Avg Stock Related CAR
in (-50, -26)
(t-statistic in
parantheses)
Avg Stock Related CAR
in (-25, -1)
(t-statistic in
parantheses)
# of observations
Negative Credit Watch
(Developed Markets)
-1.51%
(-6.91)
-1.95%
(-8.27) 2,087
Negative Credit Watch
(Emerging Markets)
-0.11%
(-0.16)
0.24%
(0.17) 302
Welch t-test for difference
in sample means
t-statistic = 1.9372
df = 363
t-statistic = 2.1822
df = 336
Positive Credit Watch
(Developed Markets)
-0.03%
(-0.12)
0.24%
(0.84) 743
Positive Credit Watch
(Emerging Markets)
-1.92
(-1.86)
-0.11
(-0.11) 88
Welch t-test for difference
in sample means
t-statistic = 1.7639
df = 100
t-statistic = 1.6396
df = 516
Total: 3,220*
Long Term Rating
Downgrade (Developed
Markets)
-2.35%
(-9.04)
-1.84%
(-7.08)
1,933
Long Term Rating
Downgrade (Emerging
Markets)
-2.79%
(-3.35)
-3.42%
(-3.69)
446
Welch t-test for difference
in sample means
t-statistic = 0.5046
df = 534
t-statistic = 1.6396
df = 516
Long Term Rating Upgrade
(Developed Markets)
0.41%
(1.96)
0.26%
(1.25)
1,183
Long Term Rating Upgrade
(Emerging Markets)
-0.35%
(-0.66)
0.11%
(0.22)
367
Welch t-test for difference
in sample means
t-statistic = 1.3236
df = 481
t-statistic = 0.3435
df = 102
Total: 3,929*
*not all obs in sample have preannouncement stock prices available
*not all obs in sample have preannouncement stock prices available
Table 12
Average Preannouncement Stock Related CAR for (-50, -26) and (-25, -1) for Entire Sample of Positive /
Negative Creditwatch and Long Term Rating Upgrades / Downgrades
Average Preannouncement Stock Related Culmulative Abnormal Return (CAR) for 2 periods, (-50, -26) & (-25, -1)
is calculated as the difference between the daily raw stock return and the concurrent local market index (as
defined in DataStream). We note that not all the observations in the previous samples have preannouncement
stock prices available.
63
11. CROSS SECTIONAL ANALYSIS OF EXCESS STOCK
RETURNS
11.1 Explanatory Variables
We estimate multivariate regressions to try to explain cross-sectional
variation in the stock-related CAR due to credit watch and bond rating changes.
Using the same methodology as Jorion (2004), Hand, Holthausen and Leftwich
(1992), etc, separate regressions are estimated for upgrades and downgrades.
The following variables are included in the regression:
1. Rating change magnitude, represented by a cardinal variable that
indicates the number of grades changed (with AAA having a score of 1,
and D having a score of 26), and the variable being new score old
score.
2. For rating changes, a dummy variable set to 1 if the rating change is not a
resolution of a prior credit watch. The criteria for resolution is that the
rating change is in the same direction as the credit watch, occurs within 1
year of the credit watch, and is the earliest rating change for that specific
64
bond after the credit watch. For credit watch, this is set to 1 if the credit
watch was informative.
3. For rating changes, a dummy variable set to 1 if the rating change moves
the bond into or out of investment grade.
4. A dummy variable set to 1 if the credit watch / rating change occurs to a
company domiciled in an developed market country, as defined in the
Morgan Stanley Capital Index classification.
5. The monthly return of the MSCI country index for the country of domicile
of the company during the month of the rating change.
6. Time lapse since the last rating change
7. Market capitalization of company during the credit watch / rating change.
8. A dummy variable set to 1 if the country enforces prohibitions against
insider trading i.e., the country has previously persecuted insider
trading.
9. An index score of Anti Director Provisions, as a proxy for shareholder
rights. This proxy of country level corporate governance provisions is
used as defined in Law and Finance, La Porta, Harvard University, 98.
The index is the sum of the following dummy variables (as defined and
reproduced from La Porta 98):
65
a. One share - one vote: Equals one if the Company Law or
Commercial Code of the country requires that ordinary shares
carry one vote per share, and zero otherwise. Equivalently, this
variable equals one when the law prohibits the existence of both
multiple-voting and non-voting ordinary shares and does not allow
firms to set a maximum number of votes per shareholder
irrespective of the number of shares she owns, and zero otherwise.
b. Proxy by mail: Equals one if the Company Law or Commercial
Code allows shareholders to mail their proxy vote to the firm, and
zero otherwise.
c. Shares not blocked before meeting: Equals one if the Company
Law or Commercial Code does not allow firms to require that
shareholders deposit their shares prior to a General Shareholders
Meeting thus preventing them from selling those shares for a
number of days, and zero otherwise.
d. Cumulative voting or proportional representation: Equals one if the
Company Law or Commercial Code allows shareholders to cast all
of their votes for one candidate standing for election to the board of
directors (cumulative voting) or if the Company Law or Commercial
66
Code allows a mechanism of proportional representation in the
board by which minority interests may name a proportional number
of directors to the board, and zero otherwise.
e. Oppressed minorities mechanism: Equals one if the Company Law
or Commercial Code grants minority shareholders either a judicial
venue to challenge the decisions of management or of the
assembly or the right to step out of the company by requiring the
company to purchase their shares when they object to certain
fundamental changes, such as mergers, assets dispositions and
changes in the articles of incorporation. The variable equals zero
otherwise. Minority shareholders are defined as those
shareholders who own 10 percent of share capital or less.
f. Preemptive rights: Equals one when the Company Law or
Commercial Code grants shareholders the first opportunity to buy
new issues of stock and this right can only be waved by a
shareholders vote, and zero otherwise.
67
11.2 Expected Signs
The stock-related CAR should be more positive depending on the number
of grades changed by the rating, and should also be smaller in absolute
magnitude for resolutions of a credit watch. Additionally, the absolute magnitude
of the rating change should also be larger for rating changes that moves the
bond across investment grades, if the bond is related to a company that is
domiciled in a developed country and if it has been longer since the last credit
watch / rating change (since the incremental amount of new information is
greater).
Market capitalization is also included in the model we hypothesize that
information availability on larger firms (by market capitalization) would be greater,
therefore reducing the informational value of assessments by credit rating
agencies. Additionally, the stock related CAR for observations in countries
where insider trading laws are enforced should be greater because there will be
less information leakage, therefore resulting in the credit event having greater
informational value.
68
Lastly, we hypothesize that the more rights shareholders have over the
firms management (as proxied by the anti director rights variable), the more
positive the reaction to all rating events should be. While rating events represent
(to some degree), the release of private information to investors, investors have
an increased capacity to take action and prevent management from acting
contrary to their interests.
69
3 Day Stock Related
CAR for Positive CW
t-statistic
3 Day Stock
Related CAR for
Negative CW
t-statistic
3 Day Stock
Related CAR for
Long Term
Upgrade
t-statistic
3 Day Stock
Related CAR for
Long Term
Downgrade
t-statistic
Dependent Variables
Intercept 0.010935686 1.199305909 -0.005486805 -0.538286329 -0.003191999 -0.724052857 -0.004611493 -0.4634405
Marketcap of Company in US
$Billions
3.25005E-12 0.56088607 6.19396E-12 0.978345961 -2.47176E-12 -0.931894101 1.10295E-11 1.8079253
Company is Domiciled in
Developed Market
-0.007038363 -1.022153037 -0.006256863 -0.805412774 0.001172058 0.430543861 -0.011118227 -1.7404421
Rating Change crosses
Investment Grade Boundary
- - - -
0.002698083 0.965929841 -0.01678223 -2.6114668
Surprised Rating Change - - - - -3.29546E-05 -0.016112677 -0.00447095 -1.0708853
Return of Local MSCI Country
Index
-0.000574349 -0.015981328 0.081633803 3.146301821 0.038899573 2.421862194 0.013415839 0.6201031
Rating Change Magnitude - - - -
0.000345362 0.56394421 1.44873E-05 0.0093546
Interval Since Last Rating
Change
6.24675E-07 0.163419916 7.54837E-06 1.687050207 -9.81837E-07 -0.517271074 1.20871E-05 2.9363613
Informative Credit Watch -0.000854063 -0.22437229 -0.000828463 -0.175801846 - - - -
Anti Director Rights -0.000400604 -0.33859153 -0.000678976 -0.509233845 0.000662591 0.9530569 -0.000563936 -0.3693984
Insider Trading Laws
Enforcement -0.002972361 -0.508222361 -0.004162292 -0.521063814 0.001265603 0.407668033 -0.003184623 -0.425067
Number of Observations 591 1,703 1,097 1,602
Adjusted R Square -0.008684848 0.00560917 0.000469852 0.01088287
R Square 0.003222055 0.009684542 0.008677673 0.016443179
Standard Error 0.041143432 0.079133616 0.030349699 0.079843862
Independent Variables
Table 13
CAR is the culmulative abnormal stock related return over the 3 day period (-1, +1). Informative Dummy, Developed Markets Dummy, Investment Grade Transition Dummy and Surprise Dummy are
dummy variable that are 1 if the credit watch is informative, if the observation is for a company domiciled in an developed markets country, if the rating change results in an investment grade transition,
and if the rating change is not foreshadowed by credit watch respectively. Contemporaneous MSCI Returns is the return on the MSCI country index for the country of domicile for the month of the rating
change. Interval Since Last Rating Change is the number of days since the last rating change event, and Rating Change Magnitude is the new rating score - the old rating score, with AAA having a score
of 1 and D having a score of 26. Anti Director rights is the index score on country level corporate governance provisions as defined in "Law and Finance", La Porta, with a higher score indicating better
corporate governance
CAR(rating change)j = B0 + B1(SURPRISE DUMMYj) + B2(DEVELOPED MARKETS DUMMYj) + B3(CONTEMPORANEOUS MSCI RETURNSj) + B4(MARKET CAPITALIZATIONj)
+ B5(INTERVAL SINCE LAST RATING CHANGEj) + B6(RATING CHANGE MAGNITUDEj) + B7(INVESTMENT GRADE TRANSITIONj) + B8(INSIDER TRADING LAWS
ENFORCEMENT) + B9 (ANTI DIRECTOR RIGHTS)
CAR(credit watch)j = B0 + B1(INFORMATIVE DUMMYj) + B2(DEVELOPED MARKETS DUMMYj) + B3(CONTEMPORANEOUS MSCI RETURNSj) + B4(MARKET
CAPITALIZATIONj) + B5(INTERVAL SINCE LAST RATING CHANGEj) + B6(INSIDER TRADING LAWS ENFORCEMENT) + B7(ANTIDIRECTOR RIGHTS)
Regression Tests on Excess Stock Returns for Companies with Credit Watch and Long Term Rating Changes by Moody's or Standard and Poor's from 1982 - 2007
70
11.3 Regression Results
The results in Table 13 shows that, for negative credit watch, there is a
more negative reaction in developed markets. In the regression used, the t-
statistic on the developed markets dummy variable is weakly significant at -
0.81, and the coefficient of -0.0063 implies that, holding all else constant, the
marginal effect of being domiciled in an emerging market decreases the stock-
related CAR from the negative credit watch by 0.63 percentage points (i.e., it is
more negative).
Additionally, the coefficient of 0.0816 on the local MSCI country index
returns implies that a 1% increase in the local contemporaneous (same month)
MSCI country index returns increases returns during a negative credit watch by
8.16% (i.e. the stock-related CAR is less negative). Lastly, although the t-
statistics on market capitalization and interval since previous rating change are
weakly significant, and the coefficients are in the correct direction, the
coefficient values are extremely small, and do not constitute a large impact on
stock-related CAR. The coefficient on market capitalization indicates that an
additional US$100 million in market capitalization only increases the stock
related CAR by 0.062 % pts (but the average market capitalization for the entire
sample is only US$128.35 million), while the coefficient on interval since
71
previous rating change indicates that an additional 100 days interval since
previous rating change increases stock related CAR by only 0.075% pts. None
of the explanatory variables for positive credit watch are significant. The
regression for positive credit watch as a whole also has no explanatory power
with a low adjusted R square.
We note that investors may factor the state of the local MSCI country
index more heavily for negative credit watch because they condition the
probability that a downgrade will materialize on the state of the broader
countrys economy. However, for positive credit watch, it is possible that since
companies more actively disseminate good news, investors are already well
informed as to the possibility of a subsequent long term rating upgrade, and
hence do not condition as heavily on the state of the broader countrys
economy.
Additionally, for long term rating downgrades, there is also a more
negative reaction in developed markets. In the regression used, the t-statistic
on the developed markets dummy variable is -1.74, and the coefficient of -
0.01111 implies that, holding all else constant, the marginal effect of being
domiciled in an emerging market decreases the stock related CAR by 1.11
72
percentage points (i.e. more negative). As expected, downgrades that cross
the investment grade band have a more negative reaction than downgrades
that do not. The t-statistic on the investment grade transition dummy is -2.61,
and the coefficient value of -0.0168 implies that, all else being equal, rating
downgrades that cross the investment grade band to non investment grade
have a stock related CAR that is more negative (-1.68 % pts). We note that
although the t-statistics on market capitalization and interval since closest rating
change are weakly significant, the coefficient values are too small to have a
notable impact on stock related CAR.
Lastly, the regression on rating upgrades demonstrate that the
contemporaneous return on the local MSCI country index significantly impacts
stock-related CAR. The t-statistic of 2.42 is significant, and the coefficient of
0.0389 implies that, holding all else constant, the marginal effect of a 1% point
increase in the MSCI country index return results in a 3.89% pt increase in
stock-related CAR.
We note that MSCI country index returns is positively related to the stock
related CAR following negative credit watch, but does not significantly affect the
stock related CAR to positive credit watch. Conversely, MSCI country index
73
returns is positively related to the stock related CAR following long term rating
upgrades, but does not significantly affect the stock related CAR to long term
rating downgrades. The former is likely due to the fact that investors use the
state of economy to determine whether a subsequent downgrade is likely, whilst
the latter may be a result of the fact that long term ratings are forward looking
with a greater time horizon than credit watches; investors may condition their
reaction to good news on whether the broader economy is also performing well,
since this could affect the companys future good prospects in the long term.
However, the fact that investors do not do this for bad news (i.e. downgrades)
could indicate that all the bad news is already factored in. This could also
explain why being domiciled in an emerging market versus developed market
does not affect stock related CAR to long-term upgrades. In short, investors
primarily weigh broader market conditions most heavily in assessing potential
positive future prospects implicit in an upgrade.
74
11.4 Discussion of Insider Trading Prohibitions
With regards to insider trading, we note that the coefficients on the
insider trading laws enforced dummy and anti director rights index variable
indicates that the enforcement of insider trading laws increases the stock
related CAR to negative CW by -0.42% (more negative) for insider trading laws,
and by -0.07% for each point on the anti director rights index. However, both
coefficients have low t-statistics which are not significant at the 2.5% level of
significance. For long term downgrades, we note an increase of stock related
CAR to downgrades by -0.32% (more negative) and -0.06% for each point on
the anti director rights index. Once again, both coefficients are not significant at
the 2.5% level of significance.
We note that one possible explanation for the significant
preannouncement abnormal returns in emerging markets compared to
developed markets (for long term downgrades), but the absence of a significant
t-statistic for the insider trading law dummy in the regression could be the
relative lack of effective enforcement of insider trading regulations in emerging
markets compared to developed markets. That is to say, it is possible that
insider trading laws may not be entirely effective in preventing insider trading,
75
especially in emerging markets. There is some evidence for this in the
literature.
In Do Insider Trading Laws Work?, European Financial Management
Journal 05, Bris shows that profits made by informed corporate insiders before
tender offer announcements increase after new insider trading laws are first
enforced. The paper finds that laws that proscribe insider trading fail to
eliminate profits made by insiders. The prohibition then shifts the supply curve
for insider trading, and therefore raises its price; insider trading therefore
becomes more profitable after laws are introduced that prohibit it. Additionally,
law enforcement also raises the possibility of monopoly profits for anyone that
can find a way to circumvent the law.
These results are supported by existing literature that compare stock
related returns to events in developed and emerging markets. Bekaert &
Harvey 02 note in a paper Research in emerging markets finance: Looking to
the Future that emerging market equity returns have higher serial correlation
than developed market returns. This serial correlation is symptomatic of
infrequent trading and slow adjustment to current information (Harvey 95,
Kawakatsu & Morey 99), and therefore, emerging market returns are less likely
76
to be impacted by company-specific news announcements than developed
market returns. The paper suggests that insider trading occurs well before the
release of information to the public.
Additionally, in a paper entitled When an Event is not an Event: The
Curious Case of an Emerging Market, Jan 2000 Journal of Financial
Economics, Bhattacharya et al showed that shares trading in the Bolsa
Mexicana de Valores (Mexican stock exchange) do not seem to react to
company news. Using a sample of Mexican corporate news announcements
from the period July 1994 through June 1997, this paper finds that there is
nothing unusual about returns, volatility of returns, volume of trade or bidask
spreads in the event window. The authors then provide evidence that suggests
that unrestricted insider trading causes prices to fully incorporate the
information before its public release.
Bekaert & Harvey 02 also point out that there is literature on stock
selection in emerging markets that suggests that simple combinations of
fundamental characteristics can be used to develop portfolios with excess
returns to the benchmark (as demonstrated in Achour et al, 99, Fama &
French, 98, Rouwenhorst, 99, etc). Bekaert & Harvey 02 conclude that the
77
preponderance of evidence therefore suggests that emerging markets are
relatively less informationally efficient than developed markets.
78
11.5 Country Breakdowns
As a robustness check, we compute the average stock related CAR for
each country for long term rating upgrades / downgrades and positive / negative
credit watch (Table 14). We find that for long term rating downgrades in
developed countries, 90.9% of the developed countries in the sample have
negative stock related CAR upon long term rating downgrade, and 61.1% of the
emerging countries in the sample have negative stock related CAR upon long
term rating downgrade. Less than 50% of developed and emerging countries in
the sample have positive stock related CAR upon long term rating upgrades.
Table 14 shows that 78.3% of developed countries in the sample exhibit
an average stock related CAR that is negative after negative credit watch.
Additionally, 61.1% of emerging countries in the sample exhibit an average
stock related CAR that is negative after negative credit watch. 54.6% of
developed countries in the sample exhibit an average stock related CAR that is
positive after positive credit watch, and 53.9% of emerging countries in the
sample exhibit an average stock related CAR that is positive after positive credit
watch.
79
Table 14
Country
Positive Credit Watch
Stock Related CAR
Negative Credit
Watch Stock
Related CAR
Long Term
Upgrade Stock
Related CAR
Long Term
Downgrade Stock
Related CAR
AUSTRALIA -0.43% -1.19% -0.03% -1.46%
AUSTRIA -0.32% 0.04% 0.42% -0.52%
BELGIUM -0.32% 0.98% -0.32% -2.07%
BRAZIL 0.00% -1.48% 0.17% -1.00%
CANADA 0.08% -0.69% 1.23% -1.56%
DENMARK 1.82% 5.65% 0.00% -0.40%
FINLAND 5.84% -2.13% -0.66% 0.19%
FRANCE -0.54% -1.74% -0.20% -1.28%
GERMANY -0.06% -0.35% -0.14% -1.03%
GREECE 3.00% 1.00% 0.02% 1.21%
HONG KONG 0.00% -1.13% 0.32% -0.59%
IRELAND 0.69% -0.63% -0.06% -3.23%
ITALY 0.14% -1.07% 0.17% -1.00%
JAPAN -0.08% -1.76% 0.15% -0.69%
NETHERLANDS -0.07% -5.15% -0.09% -3.44%
NEW ZEALAND 0.28% -1.46% -0.50% -0.01%
NORWAY -1.15% -2.55% -0.18% -6.67%
PORTUGAL 4.52% 0.46% 0.21% -0.09%
SINGAPORE N/A -4.00% -0.15% N/A
SPAIN 0.14% -0.17% -0.28% -0.57%
SWEDEN 0.01% -1.85% -0.50% -0.43%
SWITZERLAND 1.69% -9.25% 0.07% -5.06%
UNITED KINGDOM 0.36% -1.72% -0.08% -1.75%
# of countries 22 23 23 22
# of countries with
average returns in the
correct direction (i.e. > 0
for positive watch and
upgrades, and < 0 for
negative watch and
downgrades)
12 18 10 20
% of countries with
average returns in
correct direction
54.55% 78.26% 43.48% 90.91%
Listing of average stock related CAR to Long Term Rating Changes from Moody's Investor's
Service and S&P Rating Agency
The combined sample of long term rating changes is broken down by rating type (upgrades or
downgrades) and country, with the average stock related CAR of long term rating events in each country
for that rating type listed. Overall country results are considered to be in the correct direction if the rating
type is a downgrade and the average stock related CAR for all rating types for that country is negative, or
if the rating type is an upgrade and the average stock related CAR for all rating types for that country is
positive. Emerging Dummy is 1 if the country is classified as emerging in the Morgan Stanley Capital
Index
Table 14 Panel A: Developed Markets
80
Country
Positive Credit Watch
Stock Related CAR
Negative Credit
Watch Stock
Related CAR
Long Term
Upgrade Stock
Related CAR
Long Term
Downgrade Stock
Related CAR
ARGENTINA -0.30% -1.28% 0.05% -2.87%
BRAZIL 0.01% 0.77% -1.28% -0.16%
CHILE N/A -0.15% -0.53% -1.00%
CHINA 3.40% -3.44% 1.71% 0.40%
HUNGARY 0.01% 0.13% -0.33% -0.23%
INDIA 0.83% -0.70% -0.08% 5.43%
INDONESIA -0.59% 0.61% -1.88% 0.90%
ISRAEL N/A -2.33% 2.00% 3.34%
KOREA -0.47% 0.53% 1.33% 2.36%
MALAYSIA -0.55% 0.52% -0.07% -1.09%
MEXICO -0.43% -2.30% 0.05% -1.48%
PHILIPPINES 2.75% 0.28% -1.06% -0.17%
POLAND N/A -12.23% -0.68% 7.29%
RUSSIA 1.39% 0.83% -1.03% -2.33%
SOUTH AFRICA N/A -0.83% N/A 3.00%
TAIWAN N/A -3.35% 0.20% -0.79%
THAILAND 3.77% -1.62% 1.26% -1.51%
TURKEY -7.00% -21.00% 1.45% -1.00%
# of countries 13 18 17 18
# of countries with
average returns in the
correct direction (i.e. > 0
for positive watch and
upgrades, and < 0 for
negative watch and
downgrades)
7 11 8 11
# of countries with
significant results (at
2.5% level of
significance)
2 4 1 4
% of countries with
average returns in
correct direction
53.85% 61.11% 47.06% 61.11%
Table 14 Panel B: Emerging Markets
81
12. CONCLUSION AND FUTURE DIRECTIONS
We examine the informational content of being placed on Moodys and
S&Ps watch lists using a comprehensive database of credit watch placements,
and also bond rating changes for non US domiciled companies. We analyze
the informational content in 3 ways first, we examine stock related CAR from
separate samples of credit watch placements and also bond rating changes
over a 3 days (-1, +1) window centered on the actual credit watch / rating
change on day 0.
Secondly, we examine the linked samples of credit watches that are
resolved by expected rating changes, and also the unlinked samples, where the
rating changes are unexpected. Thirdly, we analyze the samples by various
partitions, include emerging / developed markets, investment grade transition /
non-investment grade transition and state of local MSCI country index.
Being placed on a credit watch list is, by itself an informative event.
Additionally, negative credit watches appear to carry a greater informational
content compared to positive credit watches this could be due to the
82
explanation offered in Goh & Ederington (1993) that companies are more
proactive in disseminating positive news compared to negative news. Long
term rating downgrades on the whole also result in a significant negative stock
related CAR. Positive credit watch and upgrades on the whole generally do not
result in significant reactions.
Reactions to negative credit watch and long term rating downgrades are
generally less pronounced (i.e. more positive) for companies domiciled in
emerging markets compared to developed markets. This could be because of
greater information leakage (e.g. through insider trading, etc) in emerging
markets that result in bad news being disseminated more rapidly than in
developed markets. Additionally, surprised long term downgrades and
informative negative credit watches all result in stock related CAR that are more
pronounced (i.e. more negative) than expected long term downgrades and
uninformative negative credit watches the same is true for both developed
and emerging markets. Reactions to credit watches and long term rating
changes also appear positively related to the contemporaneous return on the
MSCI local country index. Lastly, long term rating downgrades that result in a
transition from investment grade to non-investment grade generally exhibit a
more negative stock-related CAR.
83
Going forward, it may be interesting to analyze average stock-related
CAR at the country level and also partitioned by national regulatory
characteristics at the aggregate level. This could help to identify which specific
national regulations that mandate corporate disclosure, regulate insider trading,
and enforce corporate transparency impact the additional informational content
that bond rating changes and credit watch provide to investors.
84
13. BIBLIOGRAPHY
Baker, H, Mansi, S. A. (2002). Assessing Credit Rating Agencies by Bond
Issuers and Institutional Investors. Journal of Business Finance & Accounting,
29, 1367 - 1398.
Boot, A., Milbourn, T., & Schmeits, A. (2006). Credit Ratings as Coordination
Mechanisms. The Review of Financial Studies, 19, 81 - 118.
Dichew, I. & Piotroski, J. (2001). The Long-run Stock Returns Following Bond
Ratings Changes. Journal of Finance, 56, 173 - 203.
Ederington, L. & Goh, J. (1998). Bond Rating Agencies and Stock Analysts:
Who Knows What When? Journal of Financial and Quantitative Analysis, 33, 569
- 585.
85
Ellis, D. (1998). Different Sides of the Same Story: Investors' and Issuers' Views
of Rating Agencies'. Journal of Fixed Income, 7, 35 - 45.
Glascock, J. L., Davidson, W. N., & Henderson, G. V. (1987). Announcement
effects of Moody's bond ratings changes on equity returns. Quarterly Journal of
Business and Economics, 26, 67 - 78
Goh, J. C., & Ederington, L. H. (1993). Is a Bond Rating Downgrade Bad News,
Good News or No News for Stockholders? The Journal of Finance, 48, 2001 -
2008.
Grier, P. & Katz, S. (1976). The Differential Effects of Bond Rating Changes
Among Industrial and Public Utility Bonds by Maturity. The Journal of Business,
49, 226 - 239.
Griffin, P. A., & Sanvicente, A. Z. (1982). Common Stock Returns and Rating
Changes: A Methological Comparison. The Journal of Finance, 37, 103 - 119.
86
Hand, J., Holthausen, R., & Leftwich, R. W. (1992). The effect of bond rating
agency announcements on bond and stock prices. The Journal of Finance, 47,
733 - 752.
Hettenhouse, G., & Sartoris, W. (1976). An analysis of the informational value of
bond rating changes. Quarterly Review of Economics and Business, 16, 65 - 78
Hite, G., & Warga, A. (1997). The Effect of Bond-Rating Changes on Bond Price
Performance. Financial Analysts Journal, 53, 35 - 51.
Holthausen, R., & Leftwich, R. (1986). The effect of bond rating changes on
common stock prices. Journal of Financial Economics, 17, 57 - 89.
Jorion, P., Liu, Z., & Shi, C. (2005). Informational Effects of Regulation FD:
Evidence from Rating Agencies. Journal of Financial Economics, 76, 309 - 330.
87
Kaminsky, G. L., & Schmukler, S. L. (2002). Emerging Markets Instability: Do
Sovereign Ratings Affect Country Risk and Stock Returns? World Bank
Economic Review, 16, 171 - 195.
Katz, S. (1984). The Price and Adjustment Process of Bonds to Rating
Reclassifications: A Test of Bond Market Efficiency. The Journal of Finance, 29,
551 - 559.
Keenan, S., Fons, J., & Carty, L. (1998). An Historical Analysis of Moody's
Watchlist. Moody's Investors Service, Special Comment, October.
Kim,Y., & Nabar, S. (2007). Bankrupcy Probability Changes and the Differential
Informativeness of Bond Upgrades and Downgrades. Journal of Banking and
Finance, forthcoming.
La Porta, Lopez-de-Silanes, Shleifer, Vishny (1998). Law and Finance. NBER
Working Papers 5661.
88
Norden, L., & Weber, M. (2004). Informational Efficiency of Credit Default Swaps
and Stock Markets: The impact of Credit Rating Announcements. Journal of
Banking & Finance, 28, 2813 - 2843.
Pinches, G. E., & Singleton, J. C. (1978). The Adjustment of Stock Prices to
Bond Ratings Changes. The Journal of Finance, 33, 29 - 44.
Purda, L. (2007). Stock Market Reaction to Anticipated Versus Surprise Rating
Changes. Journal of Financial Research, 30, 301 - 320.
89
APPENDIX A: THE BEHRENS FISHER PROBLEM
The Behrens Fisher problem involves interval estimation and
hypothesis testing on the difference of means of two normally distributed
populations, when the variances of the 2 populations may not be equal. We note
that it is assumed that the 2 populations are independent.
Behrens and Fisher proposed to find the probability distribution of
=
2 1
2 1
2
1
N N
S
X X
t
s
s
s
Fisher proposed initially that the distribution of this statistic can be
approximated by ignoring random variation in the relative sides of the standard
deviations, as in:
n
s
n
s
n
s
2
2
2
1
2
1
1
1
+
90
Welch (1938) approximated the distribution by the Type III Pearson
distribution, applying this to the following number of degrees of freedom:
( )
( ) ( ) 1 1
2
2
2
1
2
1
2
2 1
+
=
n n
v
where
ni
i
i
2
=
The null hypothesis would then involve the expectation of equality,
2 1
= , so the distribution of the Behrens Fisher statistic, T, which will also
depend on the variance ratio (of both distributions) can now be approximated by
the Students t distribution with v degrees of freedom.